Volume-outcome relationship in intra-abdominal robotic-assisted surgery: a systematic review

As robotic-assisted surgery (RAS) expands to smaller centres, platforms are shared between specialities. Healthcare providers must consider case volume and mix required to maintain quality and cost-effectiveness. This can be informed, in-part, by the volume-outcome relationship. We perform a systematic review to describe the volume-outcome relationship in intra-abdominal robotic-assisted surgery to report on suggested minimum volumes standards. A literature search of Medline, NICE Evidence Search, Health Technology Assessment Database and Cochrane Library using the terms: “robot*”, “surgery”, “volume” and “outcome” was performed. The included procedures were gynecological: hysterectomy, urological: partial and radical nephrectomy, cystectomy, prostatectomy, and general surgical: colectomy, esophagectomy. Hospital and surgeon volume measures and all reported outcomes were analysed. 41 studies, including 983,149 procedures, met the inclusion criteria. Study quality was assessed using the Newcastle–Ottawa Quality Assessment Scale and the retrieved data was synthesised in a narrative review. Significant volume-outcome relationships were described in relation to key outcome measures, including operative time, complications, positive margins, lymph node yield and cost. Annual surgeon and hospital volume thresholds were described. We concluded that in centres with an annual volume of fewer than 10 cases of a given procedure, having multiple surgeons performing these procedures led to worse outcomes and, therefore, opportunities should be sought to perform other complimentary robotic procedures or undertake joint cases.


Introduction
Robotic-assisted surgery (RAS) has seen a rapid growth globally in recent years and the breadth of procedures utilising the approach is increasing with multiple surgical specialities now employing robotic surgery for a variety of procedures for benign and malignant disease. This widespread expansion of RAS means robotic platforms are no longer exclusively available to large teaching hospitals or tertiary referral centres with high case volumes and this leads healthcare providers and planners to consider the case volume and mix required by smaller centres to ensure high quality and cost effectiveness [1,2].
In general, there is evidence indicating that surgical outcomes and cost effectiveness improve with increased procedure volume, whether for a particular hospital or individual surgeon [3][4][5][6]. The volume-outcome relationship has a number of contributory factors and their relative contribution is varied and debated as is the strength of the overall relationship [7]. It is possible that the early introduction of individual robotic platforms in a geographical area may result in the redistribution of surgical volume towards a group of surgeons or an institution. Consequently, it is not clear if the reported benefits of RAS relate, in some part, to this volumeoutcome relationship.
The volume-outcome relationship is the premise for minimal volume standards that are set for specific procedures worldwide including the Volume Pledge (USA) [8,9], Scottish Quality Performance Indicators/Getting it Right First Time (UK) [10,11], and Minimum volume regulations (Germany) [12]. These thresholds are informed, at least in part, by evidence of the volume-outcome relationship for each procedure. Surgeons, their health care networks and regulatory bodies, therefore, require information on the volumeoutcome relationship in robotic surgery to consider how best to utilise platforms, especially in lower volume settings [13,14]. In addition, patients may wish to access this information to make an informed choice.
To date, research has been focussed on robotic equivalence, superiority and cost effectiveness compared to other minimally invasive or open approaches. The relationship between volume and outcome has not been addressed comprehensively across specialities utilising robotic platforms [15]. Institutions rolling out de-novo robotic programs must develop plans for surgeon training and access to robotic surgery. In the implementation phase, procedure volume and regular access is considered important to enable individuals to successfully ascend the learning curve. Beyond this point, little is known with regards to minimum procedure numbers to maintain RAS skills or what might constitute an optimum number of surgeons per system with regular RAS access to deliver a successful programme. Therefore, we sought to review the literature to determine the volume-outcome relationship in robotic-assisted intra-abdominal surgery. We conducted a systematic literature review of intra-abdominal robotic procedures performed in general surgery, gynecology and urology. Our aim was to report on whether institutional and surgeon specific volume-outcome relationships exist for RAS in these surgical subspecialities and if so, report on suggested minimum surgical volumes to maintain high quality outcomes.

Search strategy
A systematic literature search was performed using the Preferred Reporting Items for Systematic Reviews and Metaanalyses (PRISMA) statement (2020) and registered with the PROSPERO Registry [16,17]. The search used Medline, NICE Evidence Search, Health Technology Assessment Database and Cochrane Library to identify articles cited to and including the 31st October 2021. There was no start date, all records were considered. The terms used were: robot*, surgery, volume and outcome. There were no language criteria.

Study selection
Studies were included if they related to intra-abdominal robotic surgery and measured the relationship between surgeon and/or hospital volume and any perioperative outcome, including cost. Esophagectomy was included in the study, but the authors acknowledge that some approaches to this procedure are transthoracic.
Studies were excluded if they related to the learning curve, that is, they assessed outcomes from consecutive cases or related to the total number of cases a surgeon had performed (a surgeon's experience). Learning curve studies were excluded as these pertain more to the evidence base required for developing robotic training curricula rather than service design. Systematic reviews, editorial comments and letters to the editor were also excluded.

Data extraction
The identified abstracts were screened to assess if they met the inclusion criteria. 10% of the identified abstracts were independently screened by two authors (ED,NG) and this process demonstrated 100% concordance in selection of articles for inclusion.
Full-text articles were then assessed for eligibility and the data were extracted for each study and included study design, year, country, procedure type and number, cut-off values for hospital or surgeon volume and the outcomes measured. The outcomes measured were grouped into intraoperative complications, including conversion to open, postoperative complications, including length of stay, oncological outcomes (positive margins/lymph node yield) and cost.

Quality assessment
The Newcastle-Ottawa Quality Assessment Scale (NOS) was used to assess the quality of the included studies. The tool assesses the quality of nonrandomised studies and uses a star system which judges the studies on three areas: selection, comparability and outcome [18]. The criteria used and further explanation is available in Supplementary Material.

Results
The literature search identified 2956 potentially relevant records. Figure 1 illustrates how these were screened to identify 41 studies that met the inclusion criteria. The included studies are summarized in Tables 1, 2, 3, 4, 5.
Three specialties were included: General Surgery (4 studies), gynecology (6 studies) and Urology (31studies). 20 studies addressed surgeon volume and 33 studies addressed hospital volume. There were no prospective randomized controlled trials identified; studies included were retrospective observational studies.
The quality assessment of the included studies is available in Table 6. In general, most studies were of good quality, with 30 studies scoring 7 stars or more on the Newcastle-Ottawa Quality Assessment Scale. Where scores were lower, studies were scored down in the selection category for using a limited participation database or failure to include comment on data completeness and for failing to control for confounding factors in the outcome category. Confounding factors included patient demographics, comorbidities, and disease parameters.
The results of the included studies are described below by specialty and then by the outcome measured (perioperative complications, oncological outcomes and cost). The perioperative complications measured in the included studies encompassed intra-operative events, such as conversion to open, organ injury and operative time; post-operative events, such as prolonged length of stay and readmission; and functional outcomes, such as urinary continence and erectile dysfunction. The oncological outcomes measured in the included studies included positive margins, lymph node yield, disease recurrence and overall survival. All studies addressing cost originated from the United States and included colorectal procedures, radical prostatectomy and partial nephrectomy.
Three studies addressed surgeon volume. Ruiz

Identification of studies via databases and registers
Identification Screening Included Fig. 1 The PRISMA 2020 flow diagram for the identification and screening of the studies included in this systematic review to those that performed more than one and found no significant relationship with perioperative outcomes [21]. Brunes et al. did find a significant relationship between surgeon volume and perioperative outcomes, including blood loss and operative time [24]. Unger et al. pooled all robotic gynecological procedures in their analysis and demonstrated that higher surgeon volume was associated with reduced rates of conversion to an alternative procedure [20]. Two further studies addressed hospital volume. Baba et al. [22] demonstrated improved perioperative outcomes with higher volumes using a threshold of ≥ 26cases/4 years and Matsuo et al. [23], using volume as a continuous variable, described the inverse relationship with higher volumes being associated with worse outcomes. Matsuo et al. scored higher on the quality assessment, but it was noted that robotic assistance was more common in small bed capacity, non-urban centers and felt this may have confounded the results [23].
Cost was addressed in one study. Wright et al. demonstrated a significant relationship for both surgeon and hospital volume with cost [19]. The relationship was seen to a greater degree in surgeon volume and cases performed for endometrial cancer by surgeons performing more than 50 procedures a year reduced the cost by $578 compared to those performing 5-15 cases/year.
There was no assessment of any oncological outcome in gynecological procedures.

General surgery (Table 2)
Four general surgical studies were included, three colorectal and one esophagectomy. They demonstrated higher hospital and surgeon volumes were associated with fewer perioperative complications [25][26][27][28].
For the colorectal procedures, the surgeon volume thresholds used were varied: > 5cases/18 months [25] and > 30cases/year [26], but both studies had limited Table 1 Gynecology   Tables 1, 2, 3, 4, 5 summarise the 41 studies included in this systematic review. In relation to outcome measures, operative time and conversion from robotic to open surgery is recorded distinctly from other intraoperative complications as it is the sole intra-operative outcome measured in a number of studies. The significant findings described were found on a multivariate statistical analysis and where this was not performed or a relationship was only significant on univariate analysis this is indicated specifically Higher volumes were associated with Lower intra and postoperative complications and shorter operation Time Table 2 General surgery Tables 1, 2, 3, 4, 5 summarise the 41 studies included in this systematic review. In relation to outcome measures, operative time and conversion from robotic to open surgery is recorded distinctly from other intraoperative complications as it is the sole intra-operative outcome measured in a number of studies. The significant findings described were found on a multivariate statistical analysis and where this was not performed or a relationship was only significant on univariate analysis this is indicated specifically Study Although a single report, the esophagectomy study scored highly in the quality assessment. Hue et al. used a hospital volume threshold of > 9 esophagectomies /6 years to demonstrate a significant relationship with perioperative complications [28].
Higher surgeon and hospital volumes were associated with lower costs in the colorectal procedures [25,26]. In both studies the cost difference between high and low surgeon volume was over $8000 per patient.
A single study, addressing perioperative complications in radical nephrectomy, found no significant relationship with surgeon volume, but included a small number of cases (n = 573) [36].
Three studies reporting on surgeon volume in partial nephrectomy demonstrated a significant relationship with higher volume and more favorable outcomes [30][31][32]. The comparison in all three studies was between the highest volume groups (> 11, > 13 and > 30 cases/year) and the lowest volume groups (1, 1-2, < 7 cases/year), respectively.
With regards to oncological outcomes, lower positive margins rates were associated with higher volume in two [32,34] out of three studies assessing this variable [32,34,37]. Cost was assessed in two studies. Monn et al. [29] showed no association with hospital volume and Khandwala et al.
[31] demonstrated reduced costs associated with increased surgeon volume, however, this finding was not significant when hospital clustering was considered.
Tables 1, 2, 3, 4, 5 summarise the 41 studies included in this systematic review. In relation to outcome measures, operative time and conversion from robotic to open surgery is recorded distinctly from other intraoperative complications as it is the sole intra-operative outcome measured in a number of studies. The significant findings described were found on a multivariate statistical analysis and where this was not performed or a relationship was only significant on univariate analysis this is indicated specifically  [41]. In terms of oncological outcomes, increased lymph node yield [38,40] and lower rates of positive margins [40] were associated with increasing hospital volume.
Cost was not assessed in any of the five studies.
With regards to hospital volume, a significant association was seen with increasing volume and better perioperative outcomes in six of the nine studies assessing these outcomes [42,45,48,49,51,52,56,58,59]. In those studies demonstrating significance, the annual thresholds varied from 30 cases/year [42] to 145 cases/year [45].   Urinary continence and erectile function   Tables 1, 2, 3, 4, 5 summarise the 41 studies included in this systematic review. In relation to outcome measures, operative time and conversion from robotic to open surgery is recorded distinctly from other intraoperative complications as it is the sole intra-operative outcome measured in a number of studies. The significant findings described were found on a multivariate statistical analysis and where this was not performed or a relationship was only significant on univariate analysis this is indicated specifically Gershman et al. modelling increasing hospital volume as a nonlinear continuous variable showed it was independently associated with improved rates of perioperative complications up to approximately 100 cases/year, beyond which there appeared to be marginal improvement [52]. Higher hospital and surgeon volumes were associated with shorter operative times [46,53,58].
Cost was addressed in six studies and found to have a significant association with volume in all [42,43,45,50,52,54]. Two studies addressed the association with both surgeon and hospital volume. Cole et al. [50] found that nearly a third of the variation in robot-assisted radical prostatectomy cost was attributable to hospital characteristics, whereas, in contrast, Hyams et al. [43] demonstrated an association with lower cost and higher surgeon, but not hospital volume. However, the latter study [43] used a smaller number of cases from a single state and used cases from earlier in the adoption of robotic surgery.

Discussion
Over the last four decades, the relationship between quantity and quality in surgery has been demonstrated [3,4]. Therefore, it is not unsurprising that the studies identified in this review have demonstrated a number of volume-outcome relationships in robotic-assisted procedures. However, it is not possible to collate the evidence through a meta-analysis as the included studies use over-lapping patient cohorts and volume is categorised using different thresholds (e.g., in tertiles, quintiles or as a continuous variable). It is, therefore, challenging to propose minimal volume standards to assist service design and patient choice.
Minimal volume standards need to consider the relative merit of the outcomes that are related to volume. This analysis focused on perioperative outcomes, oncological outcomes and costs and the volume relationship described for each varied. For example, Xia et al. showed that higher annual volumes were required to improve oncological outcomes as a lower positive margin rate in radical prostatectomy was associated with a hospital volume over 72 cases/ year compared to 45 cases/year for lower rates of perioperative complications [56].
Also, to further confound the derivation of standards, surgeon volume could be considered cumulatively across related procedures. Hayn et al. showed operative time and blood loss was reduced when surgeons performing robotic-assisted radical cystectomy also had robotic-assisted prostatectomy experience [60]. Unger et al.'s analysis considered this variable and pooled a surgeon's volume from all gynecology procedures in their analysis [20]. This is an important consideration when defining minimum volume standards as, in some regions, a given surgeon may perform multiple different procedures or operate in different geographical sites using the same robotic platform.
In addition, volume-outcome findings, and hence any derived minimal volume standards, may not be generalizable from one healthcare system to another. Outcomes relating to the delivery of care such as length of stay and cost, are likely to be highly dependent on the local context but also clinical management, such as transfusion practices [61,62]. A high proportion of the identified literature relates to data gathered in the United States and all studies addressing cost originated from the United States and, therefore, results may not be generalizable.
The authors have summarized the significant annual case volume thresholds identified for clinical outcomes in the included studies ( Table 7). The ranges described could be used to guide recommendations for minimal volume stands, but are limited by the data available, with a notable paucity of data for the specialties other than urology, and the caveats we have discussed.
With regards to the cost effectiveness, ten of the 11 reports addressing this found a significant relationship. However, it may be more important to consider the cost effectiveness of the robotic platform as a whole, rather than by procedure. Feldstein et al. describes instrument variation leading to increased costs as well as a scheduling threshold of 250-325 total annual cases for a single robotic platform despite a theoretical capacity of 780 cases/year [63]. Over this threshold, access to a platform may become challenging and limit productivity. This approach to assessing the volume outcome highlights the importance Table 7 The range of volume thresholds for significant improvement in all clinical outcomes The thresholds are given in cases/per year and the total number of studies contributing to the range are included in brackets. Cost and operative time alone have been excluded as outcomes, all other outcomes have been included in the summary. Where no data were available '-' is entered of standardisation, oversight and co-ordination within a robotic programme. The main limitation of this review is the absence of a meta-analysis of the collated studies. This is due to the variation in the threshold used and outcome definitions. It is important that future studies should consider how their data may be collated with other sources and/or more widely applied. The approaches using volume as a continuous variable and looking for a plateau in the volumeoutcome relationship may be most useful.
Another limitation is that 20 (49%) of the studies included use data from more than 10 years ago. The volume-outcome relationship is likely to weaken as a procedure becomes more established and outcomes improve in general over time [64,65]. This is in part due to the adherence to an evidence based surgical technique and perioperative care that is the result of iterative quality improvement [66]. Therefore, irrespective of the utility of minimal volume standards, there is a role in every setting for an outcomes-based quality improvement programme.
In general, for intra-abdominal robotic-assisted surgery, high volumes are associated with better outcomes and increased cost effectiveness. It is reasonable to recommend trained robotic surgeons should maintain regular access to robotic cases, and seek, where possible, to optimise their volume. In centres with an annual volume of fewer than 10 cases of a given procedure, having multiple surgeons performing these procedures is not desirable and where it is necessary, opportunities should be sought to perform other complimentary robotic procedures or undertake joint cases.
As robotic platforms become more widely available, volume linked outcomes are balanced with equity of access, especially across less densely populated areas. Defining evidence based minimal volume standards will be helpful to inform this balance but, as we have described, it is challenging. Going forward, it will be important to create opportunities for collaborative data collection with unified volume and outcomes measures to assure and improve quality in robotic-assisted surgery.
Author contributions All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by EKD and NJG. The first draft of the manuscript was written by EKD and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.
Funding The authors have not disclosed any funding.